I have watched the fortunes of the semiconductor manufacturers improve with wafer size while their support vendors have been on a sort of increasing financial diet. US mask vendors in particular were beggered by playing vendors against one another leaving none with the funding levels for R&D that the semiconductor houses enjoyed. The result was predictable.
In the 150/200mm transition, efficiency and yield improved for the fabs while the tool vendors found themselves selling fewer tools with which to cover their development costs. They learned the lesson.
As long as there is sufficient financial incentive for S/C tool and support vendors to thrive, rather than just survive, then 450mm will happen in a timely fashion. Otherwise vendors will predictably drag their feet until the semiconductor companies come forward to share in the expense of the requisite R&D funding.
Several years ago, a major customer requested us to gear up a fab in advance for the the production of their new device still in design. It was an expensive request with a lot of risk if they did not follow through with the orders. A clever gentleman within our organization struck an agreement where the customer would buy all the required expansion equipment which we would install then build the devices for them at an adjusted price. If the customer had defaulted, we were protected and they were out a significant sum.
Hard investment dollars by the customer(s) is best grease to get things unstuck.
450mm would be about 2 years earlier but the economic downturn slowed everything down and pushed out any 450mm plans. This is now mainly a question of money. The equipment suppliers paid mainly the prize for the 300mm transition which began back in 1996 to 2000. Who is paying now? Well the ones who have the money: Intel, TSMC, Samsung, maybe two more. This will trigger a domino effect which means more companies feel encouraged to jump on the train. However, these fabs and tools will be very expensive, so I see JVs and Alliance first.
I did a blog on the 450mm Semiconductor Manufacturing Debate, which was based on one of the longest LinkedIn debates I have seen.
450mm will happen, believe it!
Blog Doing Math in FPGAs Tom Burke 2 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...